Figure 4.Upper panels: Scalp distributions of the differences for depressed individuals (left) and for healthy controls (right) are depicted. Lower panels: Grand-average ERP waveforms were pooled across electrode sites Pz, Cz, CPz, C1, and C2 during imagery for depression (left) and control (right).

4.Discussion

In this study, we investigated the neural response to emotional imagery in depressed individuals and healthy controls when they imagine different emotional images using an encoding-imagery paradigm. The behavioral results showed that depressed individuals scored lower on valence ratings for sad and neutral imagery compared to healthy controls. Furthermore, the ERP findings revealed that depressed participants exhibited enhanced LPP responses to sad imagery relative to happy imagery, whereas healthy controls showed greater LPP response to happy imagery relative to sad imagery. Notably, depressed individuals exhibited enhanced LPP responses to sad imagery compared to healthy controls in both middle and late time windows, but not during the early time window.
Our findings indicate that there are differential LPP responses to emotional imagery over centroparietal sites between depressed individuals and healthy controls, which are broadly consistent with previously observed LPP responses in depression (Auerbach et al., 2015; Dainer-Best et al., 2017; Speed et al., 2020). Specifically, depressed individuals exhibited enhanced LPP responses to sad imagery relative to happy imagery, whereas healthy controls demonstrated the opposite pattern. Namely, depressed individuals exhibit similar pattern of neural responses to sad imagery as they do to negative autobiographical memories, and self-referenced negative words. These findings may be related to the sustained attentional engagement to negative stimuli in depressed individuals (Auerbach et al., 2015; Dainer-Best et al., 2017), and may be partially explained by the potent emotional impact of imagery and its close association with emotional memory (Holmes et al., 2009, 2016). Notably, no significant difference was found in LPP responses between emotional (happy and sad) and neutral imagery, which probably because the LPP response to neutral images containing people were comparable to emotional images (Ferri et al., 2012; Weinberg & Hajcak, 2010).
Critically, depressed participants showed larger LPP responses to sad imagery compared to healthy controls, especially during the middle and late time windows. These findings support the presence of mood-congruent imagery biases in depressed individuals (LeMoult & Gotlib, 2019). According to Beck’s cognitive model of depression, individuals with depression have mood-congruent schemas that lead depressed individuals to exhibit negative information-processing biases and distort the processing of emotional stimuli, leading to enhanced reactivity (Beck, 1967). Additionally, the negative bias for sad imagery in depressed individuals was mainly observed during the middle and late time windows, suggesting that depression-related abnormalities in emotional processing primarily occur more during the top-down attentional capture phase rather than automatic attentional capture phase of emotional information. Namely, depression risk is associated with later and more elaborate processing of negative emotional information (Speed et al., 2016). Benau et al. (2019) also reported group differences in LPP were specific to the middle time window. The possible reason is that the earlier LPP reflects automatic attentional capture of salient information, and the later LPP is more influenced by top-down attentional capture (Hajcak et al., 2010; Olofsson et al., 2008). In summary, our findings indicate that sad imagery captures more cognitive resources during later stages of information processing in depressed individuals.
However, our results contrary to the study conducted by Bauer and MacNamara (2021), where they observed a blunted LPP to negative imagery in depressed individuals. These discrepancies in the LPP to valence may be attributed to variations in factors such as age of depression onset, participant characteristics, stimulus properties, and task types (Benau et al., 2019; Grunewald et al., 2019; Weinberg et al., 2016). Bauer and MacNamara’s (2021) study involved participants with complex internalizing psychopathology and utilized general negative imagery as stimuli, which may have influenced the neural response. Depressed individuals process negative stimuli, particularly sad ones, better and more accurately (Delle-Vigne et al., 2014). In our study, we utilized sad and happy images that contain people as emotional stimuli instead of general positive (i.e., erotic, food or flower images) and negative stimuli (i.e., sad or threatening scenes). Consequently, we did not observe a main effect between depressed individuals and healthy controls in LPP amplitudes for happy and neutral imagery, but only for sad imagery. The differences likely reflect specific processing biases toward sad imagery rather than a broad bias in emotional information processing. Collectively, our findings suggest that an enhanced LPP response to sad imagery may serve as a potential biomarker of depression risk.
In line with the ERP results, depressed individuals rated sad and neutral imagery more negatively compared to healthy controls, while their ratings for happy imagery were comparable. Our behavioral results are consistent with previous studies showing a bias toward endorsing and recall more negative stimuli in depression (Benau et al., 2019; Dainer-Best et al., 2017; Speed et al., 2016). The finding provides behavioral evidence for depressed participants experiencing more negative imagery (Holmes et al., 2016; Weßlau & Steil, 2014), supporting the presence of mood-congruent biases in depression. Overall, the tendency toward a negativity bias in imagery valence ratings observed in depressed individuals may serve as a potential cognitive feature associated with vulnerability to depression.
Caution should be exercised when interpreting the present results due to several limitations. Firstly, the relatively small sample size limited the interpretability of effect sizes, and the use of homogeneous college student limited the generalizability of these findings. Future studies should employ larger and more diverse populations to enhance statistical power and generalize to broader populations. Second, our study compiled a new set of pure emotion materials to ensure idiographic emotional stimuli for happy, sad, and neutral images. However, these images were used for the first time in the present study, replication in future studies is necessary to establish their reliability and validity. Finally, although our study identified enhanced LPP activity to sad imagery as a potential risk indicator for depression, it did not include longitudinal follow-up to determine whether those participants with enhanced LPP to sad imagery subsequently develop more severe depressive symptoms.
In summary, this study contributes novel behavioral and electrophysiological evidence regarding emotional imagery differences between depressed individuals and healthy controls. Both behavioral and ERP results support the presence of a mood-congruent processing bias in depressed individuals. Additionally, sad imagery bias is evident in the middle and late time windows, indicating a greater engagement of top-down attentional resources in depressed individuals. These findings shed light on the cognitive and neural processes underlying emotional imagery in depression.

Reference

Auerbach, R. P., Stanton, C. H., Proudfit, G. H., et al. (2015). Self-referential processing in depressed adolescents: A high-density event-related potential study. Journal of Abnormal Psychology , 124(2), 233-245. https://doi.org/10.1037/abn0000023.
Bauer, E. A., & MacNamara, A. (2021). Comorbid anxiety and depression: Opposing effects on the electrocortical processing of negative imagery in a focal fear sample. Depression and Anxiety , 38(8), 690-699.